Trailing-Edge
-
PDP-10 Archives
-
tops20tools_v6_9-jan-86_dumper
-
tools/sed2/cobb.let
There are 2 other files named cobb.let in the archive. Click here to see a list.
December 28, 1983
Dan Cobb
LAN SWE, MR03-1/K20
2 Iron Way, Box 1003
Marlboro, MA 01752
Dear Dan,
Thank you for your letter regarding SED. It looks like the
VAX version of SED is becoming fairly widespread, especially
among former TOPS-10 and TOPS-20 people. I wish I had more
positive information to give you about the questions you
asked, but I don't have much more information at this time.
I will try to address the questions you asked in your
letter.
Regarding the XON/XOFF problem, I don't yet have a solution,
but it looks like one may be on the horizon. I have
investigated the problem and come to the conclusion that it
is a problem with VMS itself. This was corroborated by Dave
Weaver, another former TOPS user now using the VAX version
of SED. The problem is in SED's use of the PASSALL mode for
terminal input. I use the temporary PASSALL mode for input
so that VMS won't do anything with the terminal screen.
Apparently when VMS has the terminal line in passall mode,
it ignores any incoming XONs and XOFFs. This means that if
the terminal sends an XOFF while VMS is doing the output,
the output can be stopped. By the time the terminal sends
the corresponding XON, the program is back waiting for input
and VMS then ignores the XON, thus hanging the process.
There are apparently some changes coming in VMS version 4.0
that should eliminate this problem. Dave Weaver said that
there will be a PASSTHROUGH mode available with 4.0 that
will pass all characters through to the program while
observing the XON/XOFF protocol. He said he might have a
chance to try it on a system running a field test version of
4.0 and will let me know the results. If that will fix the
problem, it surely would make a lot of people happy.
Regarding VT102 support in SED, that should be very easy to
implement. It just involves adding another entry to the list
of valid terminal types; for now the existing VT100 input
and output tables could be used, since the VT102 can do all
things the VT100 can do. Eventually it would be useful to
have a separate output table, since the VT102 can also do
character inserts and deletes (I think).
I have not yet investigated having DCL parse the SED command
line, although I wouldn't rule that out as a future
possibility. I just haven't had time to look into that yet.
I don't understand why the SED.EXE file on the tape had no
definition for the VT100 up-arrow key; that definition has
been in SED from the start. There could possibly have been a
problem with the tape.
I will put your letter in my SED file, and I can notify you
when I have a new version of SED ready to distribute. I
don't anticipate sending a new version until VMS 4.0 comes
out, assuming I can solve the XON/XOFF problem at that time.
Thanks for writing; if you have suggestions for improving
SED, please don't hesitate to write.
Sincerely,
Paul Malquist